search

TobyPDE / FRRN

Full Resolution Residual Networks for Semantic Image Segmentation
MIT License
278 stars 93 forks source link

Dependency Version Requirements? #23

Open CJMenart opened 7 years ago

CJMenart commented 7 years ago

Hello!

I know your code is still somewhat in change, but would it be possible for you to list the version numbers of Theano and Lasagna you're using in the readme? I had to bounce around a bit. It appears that the code requires Lasagna 0.2, and Theano 0.9, but you're using their 'old' gpu backend?

Also, is there any way to run the code without cuDNN? If not, it's fine; I should really break down and buy a better GPU, but I thought I'd ask :)

TobyPDE commented 7 years ago

Hey,

Yes, you need lasange 0.2 and theano 0.9. I will add version checks to the dependency checker.

The code works with the new and the old GPU backend. I tested this before uploading the refactor.

Unfortunately, the current implementation relies on cuDNN for the spatial log softmax implementation. You could replace the operation by a theano graph that reshapes the tensor to apply the built in softmax op. However, I'd suspect that this would have a negative impact on the performance.